Reduce memory usage when SC materials differ only in orientation

[tkittel]

I just came across this new interesting publication: https://www.osti.gov/pages/servlets/purl/3002380 / https://doi.org/10.1016/j.nima.2025.171079

@KBGrammer first of all nice work! And then to my question: Is it correctly understood that the memory concerns mentioned in the conclusions of the publication relate to the scenario of having the exact same single crystal material appearing multiple times in a simulation, with just the orientation being different?

If so, then I guess we could easily fix this on the NCrystal side: Just always put the crystal in a particular orientation behind the scenes, and then simply have a particular additional rotation associated with each user-facing instance of the material.

[KBGrammer]

Yes, that's what I meant by the memory concern. I ran into that issue with a large focusing array of graphite crystals, for example.

I was able to get around it on the MCNP side by attaching a transformation to each cell and using the same material. It works, but an NCrystal solution would prevent the same issue in other codes and keep the same behavior in all of them.

[tkittel]

Ok, thanks for the fast feedback!

I am a bit busy right now trying to wrap up the next NCrystal release, but after that this seems like a reasonably low-hanging fruit to try to get done.